ENERGY MONITORING FOR EARTH OBSERVATION BIG DATA PROCESSING: A GAP ANALYSIS

—The processing of Earth observation big data (EOBD) consumes significant amounts of energy. Current EOBD processing practices largely neglect energy efficiency, and reliable information on energy consumption is often missing. To enable better energy monitoring for EOBD processing, this paper identifies the gaps by analyzing various data sources for energy-related measurements, including hardware-level power metrics, operating system-level resource utilization data, and performance statistics from distributed processing frameworks. Our analysis reveals several key gaps in current energy monitoring practices, such as variations in sensor sampling intervals, lack of transparency in virtualized cloud environments, hidden energy costs in distributed processing operations, absence of standardized benchmarking frameworks for energy consumption, and how to accurately attribute energy consumption to individual processes. To address these challenges, it is necessary to develop a comprehensive energy monitoring and profiling mechanism of energy consumption and perform extensive benchmarking with real-world EOBD tasks. These will allow users to evaluate the energy efficiency of their EOBD processing workflows, helping them optimize for reduced energy consumption.